mobilevit-x-small_alpha0.7_temp5.0

This model is a fine-tuned version of c14kevincardenas/ClimBEiTv2 on the c14kevincardenas/beta_caller_284_person_crop_seq_withlimb_2 dataset. It achieves the following results on the evaluation set:

• Loss: 0.8373
• Accuracy: 0.6621

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 5e-05
• train_batch_size: 64
• eval_batch_size: 64
• seed: 42
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_ratio: 0.1
• num_epochs: 20

Training results

Training Loss	Epoch	Step	Validation Loss	Accuracy
1.1201	1.0	90	1.3907	0.2915
1.0793	2.0	180	1.3214	0.3775
0.9679	3.0	270	1.1766	0.4644
0.844	4.0	360	1.0511	0.5316
0.745	5.0	450	0.9649	0.5810
0.7137	6.0	540	0.9291	0.5800
0.6625	7.0	630	0.8924	0.6235
0.6515	8.0	720	0.9113	0.6018
0.604	9.0	810	0.8726	0.6324
0.6031	10.0	900	0.8651	0.6472
0.5679	11.0	990	0.8491	0.6532
0.5615	12.0	1080	0.8680	0.6324
0.5364	13.0	1170	0.8812	0.6255
0.507	14.0	1260	0.8634	0.6334
0.5122	15.0	1350	0.8725	0.6383
0.501	16.0	1440	0.8544	0.6502
0.4927	17.0	1530	0.8533	0.6502
0.4891	18.0	1620	0.8573	0.6482
0.4767	19.0	1710	0.8373	0.6621
0.4903	20.0	1800	0.8546	0.6472

Framework versions

• Transformers 4.45.2
• Pytorch 2.5.0+cu124
• Datasets 3.0.1
• Tokenizers 0.20.1